Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Apr 09 2018 07:31
    @Arachnid banned @JennyJennywren_twitter
Danny Ryan
@djrtwo
the beacon chain is the "main chain" in terms of the validator sharded landscape
Paul Hauner
@paulhauner
Thank you @djrtwo
Collin K Cusce
@collincusce
Danny Ryan
@djrtwo
Hm. Might have overloaded the server. Will move it to a gist spo
Soon
Or fix seger
Server. Damn phone
Danny Ryan
@djrtwo
Should be fixed @collincusce
Collin K Cusce
@collincusce
thanks @djrtwo :-D
Andrew
@vtbradle_twitter
Does anyone have more detailed info on the inactivity leak and how it works?
Andrew
@vtbradle_twitter
@djtwo is there a fleshed our spec for the inactivity leak mechanism beyond what’s in the latest Casper paper?
Gernot Pokorny
@gernotpokorny
I write the Casper paper. Does this mean with Casper synced node A can claim state X is valid and synced not B can claim state Y is valid and non-synced node C does not not at all which state is valid and has to believe either A or B without having any proof of what is valid, because they cannot deliver proofs? Vitalik refers to that as weak subjectivity.
Lars Pensjö
@larspensjo
I think that is an issue at a long range attack. If this is the case, the Ethereum is currently undergoing an attack. If you have a node that has been offline for a long time, you can be fooled by such an attack. In practice, I don't think it is a problem. After syncing the node, just do a check with some external service you trust that you are on the right chain.
Gernot Pokorny
@gernotpokorny
Of course nobody knows how this will play out, there's a decent chance that this ends up in chaos, liike mentioned above.
Everybody can claim something and nobody can proof
As soon as there are two states new nodes have no idea what is true and nobody can proof them something. So they just trust some community. This is totally different then PoW/Bitcoin
I think we can do better...
Furthermore convergence algorithms are not exactly centralized but are not really tastey in case they really play out in non-chaos.
Lars Pensjö
@larspensjo

I don't see why there would be any chaos. As explained in the POS FAQ regarding weak subjectivity:

Note that all of this is a problem only in the very limited case where a majority of previous stakeholders from some point in time collude to attack the network and create an alternate chain; most of the time we expect there will only be one canonical chain to choose from.

Gernot Pokorny
@gernotpokorny
I do not refer with chaos to the long range attack. I refer to the custom logic within each node. Each node has to have it's own custom logic for the convergence.
Most likely the when you "snapshot" a history of checkpoints and analyze the logic within each client at all times (which you of course cannot do) you would find out that what happaned there is not so tastey, but cannot be referred to as exactly centralized.
Gernot Pokorny
@gernotpokorny
In case it does not end up in chaos (chaos would be that they simply cannot agree on a block and get constantly slashed or do vote, because their logic says wait for node A to vote first and follow node A's vote, node B waits for node A etc etc there can be anything to get stuck and lets say the last block before the checkpoint then they may vote or not and get slashed or not, because they will not magically all vote the same within the last block before the checkpoint)
You see there's much potential for chaos in the convergence and much potential for not reaching consensus, because there are no proofs to new nodes. It's hard to predict how this all will play out, but it does not feel very good from a theoretical standpoint.
Of course they have put much work into it and wrote a paper and proofs etc but it seems not to be the best solution, but that's just my opinion.
Lars Pensjö
@larspensjo

Each node has to have it's own custom logic for the convergence.

As long as everyone use the same node logic, you are ensured you will maximize profits. Anyone changing the internal logic will be punished.

chaos would be that they simply cannot agree on a block

The protocol is defined in such a way that it is possible to prove that nodes will converge on the same block if they are honest.

Gernot Pokorny
@gernotpokorny
As long as everyone use the same node logic, you are ensured you will maximize profits. Anyone changing the internal logic will be punished.
and how should this logic look like?

The protocol is defined in such a way that it is possible to prove that nodes will converge on the same block if they are honest.

where do you find the description of that in the paper?

Lars Pensjö
@larspensjo

For Casper CBC, you can read about the "consensus safety proof" at https://github.com/ethereum/cbc-casper/wiki/FAQ. It is a proof that protocol-following nodes will make the same correct decision (in some context).

For Casper FFG, see the safety proof at https://ethresear.ch/t/casper-ffg-with-one-message-type-and-simpler-fork-choice-rule/103

Lars Pensjö
@larspensjo

and how should this logic look like?

This is still being researched. I understand the most promising solution currently is something called "Shasper", because it combines sharding with Casper. See https://www.trustnodes.com/2018/07/28/shasper-specification-almost-ready-just-100-lines-finalize-says-eth-dev

Lars Pensjö
@larspensjo
Maybe the latest one can be found at https://ethresear.ch/t/beacon-chain-casper-ffg-rpj-mini-spec/2760, where you can read more about consensus safety.
Gernot Pokorny
@gernotpokorny

I think that is an issue at a long range attack. If this is the case, the Ethereum is currently undergoing an attack. If you have a node that has been offline for a long time, you can be fooled by such an attack. In practice, I don't think it is a problem. After syncing the node, just do a check with some external service you trust that you are on the right chain.

What you talk about? In PoW you can verify everything as a new node

Andrew
@vtbradle_twitter
@vbuterin can ugh take a look at Matt Green’s question? @https://twitter.com/matthew_d_green/status/1037460328190234625?s=21
Can you*
Andrew
@vtbradle_twitter
Thanks Danny
Andrew
@vtbradle_twitter
@djrtwo listened to you on HashingItOut earlier, nice job, you have time for a quick chat with a few potential stakers? We have some questions on the 32 ETH = 1C resource scaling model. If not mind if I post our questions or reach out on Twitter?
Danny Ryan
@djrtwo
Yeah, I just answered some questions about it on twitter. happy to explain further
Andrew
@vtbradle_twitter
@djrtwo Shot you a link in DM
Andrew
@vtbradle_twitter
Ryan Sean Adams is with me, saw the conversation you guys were having on twitter
Skaag Argonius
@skaag
This has probably been asked before, but can someone become a staker right now?
Skaag Argonius
@skaag
And is the staking amount 1500 ETH, or 32 ETH? :-)
Andrew
@vtbradle_twitter
Not until Casper goes live, looking like 32ETH at the moment
Paul Hauner
@paulhauner
1500 ETH was from EIP-1011 (Casper in a smart contract on the present Ethereum chain). That has been deprecated in favour of “Eth 2.0” and the new number is 32 ETH.
Skaag Argonius
@skaag
awesome
I'm looking forward to participating in this
flipchan
@flipchan
was an update recently pushed to rinkeby testnet? my solidity code doesnt work the way it use to 2 weeks ago
Genysys
@Genysys
@flipchan this isnt the channel for that
and yes rinkeby has forked constantinople changes